Discord Backup Bot

Archives a single Discord server's text channels in full fidelity: messages, attachments, embeds, reactions, reply/thread structure, edits, and deletions. Stores everything in a local SQLite database with attachment files on disk. Resumes cleanly after crashes or restarts.

Setup

1. Create a Discord bot

Open https://discord.com/developers/applications and create a new application.
Under Bot, click Reset Token and copy the token somewhere safe.
Under Privileged Gateway Intents, enable:
- Message Content Intent (required)
Under OAuth2 → URL Generator:
- Scopes: bot
- Bot Permissions: View Channels, Read Message History
Open the generated URL and add the bot to your server.

2. Get the IDs

Enable Developer Mode in Discord: User Settings → Advanced → Developer Mode.

Right-click your server icon → Copy Server ID (this is guildId). Right-click each channel you want backed up → Copy Channel ID.

3. Install and configure

Requires Node.js 22.5 or newer (uses the built-in node:sqlite module — no native compilation).

npm install

Copy config.example.json to config.json and fill in:

{
  "token": "your bot token",
  "guildId": "your server id",
  "channelIds": ["channel id 1", "channel id 2"],
  "dataDir": "./data",
  "attachmentConcurrency": 3
}

4. Run

npm start

The bot will log in, attach its live listener, start downloading attachments in the background, and backfill each configured channel (walking backward through old messages and forward to catch anything missed during downtime). Leave it running 24/7. Stop with Ctrl+C — next start resumes from the last checkpoint.

What gets captured

Messages (content, author, timestamps, reply target)
Edits — full edit history from when the bot is present
Deletions — soft-delete flag with timestamp
Attachments — downloaded to data/attachments/<channel_id>/<message_id>/<filename>
Embeds — link previews, bot embeds (stored as JSON)
Reactions — per user, with add/remove timestamps
Threads — active and public archived; new threads via live events
Author info — username, display name, avatar URL snapshot

Limitations

Pre-bot edits and deletions are unrecoverable. Discord's API does not expose edit/deletion history — the bot only sees them live.
Reactions removed before the bot joined are lost. Reactions still present at backfill time are captured.
Private archived threads are not backfilled (requires extra permissions). Private threads created while the bot is live are captured.
Signed attachment URLs expire after ~24 hours. If the bot is offline that long after a message is posted, its attachment URL may 403/410 before the downloader can fetch it — metadata is preserved but the file is unrecoverable.
Gateway reconnects that don't trigger a fresh ready don't re-run backfill. Restart the bot to catch up after long disconnects.

Storage

data/
├── backup.db          # SQLite (WAL mode; -shm / -wal files appear while running)
└── attachments/
    └── <channel_id>/<message_id>/<attachment_id>/<filename>

Inspecting the archive

sqlite3 data/backup.db

Useful queries:

-- Messages per channel (including threads)
SELECT c.name, COUNT(*) AS count
FROM messages m JOIN channels c ON c.id = m.channel_id
GROUP BY c.id ORDER BY count DESC;

-- Backfill progress
SELECT name, is_thread, backfill_complete,
       oldest_fetched_id, newest_fetched_id
FROM channels;

-- Pending / failed attachments
SELECT
  SUM(CASE WHEN downloaded = 1 THEN 1 ELSE 0 END) AS done,
  SUM(CASE WHEN downloaded = 0 AND download_failed = 0 THEN 1 ELSE 0 END) AS pending,
  SUM(download_failed) AS failed
FROM attachments;

-- Messages with edit history
SELECT m.id, m.content, COUNT(e.message_id) AS edits
FROM messages m LEFT JOIN message_edits e ON e.message_id = m.id
GROUP BY m.id HAVING edits > 0;

-- Deleted messages
SELECT id, author_id, content, deleted_at FROM messages
WHERE deleted_at IS NOT NULL ORDER BY deleted_at DESC LIMIT 20;